Increasingly, educators and parents want to know more than just how a student's performance compares with that of other students: they ask, "What level of performance does a score represent?" and "How good is good enough?"
To be able to adequately answer these questions, criterion-referenced interpretations are required. A criterion-referenced interpretation of a test score compares the specific knowledge and skills measured by the test to the student's proficiency with the same knowledge and skills. Criterion-referenced scores have meaning in terms of what the student knows or can do, rather than in relation to the scores produced by some external reference (or norm) group (norm-referenced interpretation). Criterion-referenced standards describe what students should be able to do at a specific level of performance.
When performance standards are developed, typically a group of experts in the field (e.g., curriculum specialists, test developers, business leaders) evaluate the test items and determine what level of mastery is necessary to be at each performance level in the content area. Increasingly, four or five performance levels are established to describe a student’s level of mastery: below basic, basic, proficient, advanced (and perhaps one additional level). The “proficient” level is often designated as “passing,” or showing adequate mastery of the content area. Performance standards reflect the judgment of the persons setting the standards and may change over time as higher (or lower) standards are set. To see examples of how standards in non-educational contexts may vary, click here.
Each test developer or state department determines its own performance standards. As a result, the Lexile or Quantile measure indicating a passing score in one state may be different than the measure that indicates passing in another state or on another test linked with the Lexile or Quantile Frameworks.
The Lexile Framework for Reading provides a context for examining performance standards from two perspectives—reader-based standards and text-based standards. Reader-based standards are determined by examining the skills and knowledge of students identified as being at the requisite level (the examinee-centered method) or by examining the test items and defining what level of skills and knowledge the student must have to be at the requisite level (the task-centered method). A cut score is established that differentiates between students who have the desired level of skills and knowledge to be considered as meeting the standard and those who do not. Text-based standards are determined by specifying those texts that students with a certain level of skills and knowledge (for example, a high school graduate) should be able to read with a specified level of comprehension. A cut score is established that reflects this level of ability and is then annotated with benchmark texts descriptive of the standard.
A norm-referenced interpretation of a test score expresses how a student performed on the test compared to other students of the same age or grade. Norm-referenced interpretations of reading test results, however, do not provide any information about what a student can or cannot read or how well a student understands mathematics. Norm-referenced interpretations simply compare student performances without regard to specific content. For accountability purposes, percentiles, normal curve equivalents (NCEs), and stanines are often used to report test results when making comparisons (norm-referenced interpretations). For a comparison of these measures, refer to the figure below and the following descriptions.
Although norm-referenced interpretations provide useful information about how a student’s score compares to the scores of a comparison group (e.g., same age or grade students), norm-referenced interpretations do not tell us whether a student has mastered the material for a particular course or grade. Norm-referenced standards do not describe what students can do at a specific level of performance. A criterion-referenced test and interpretations as well as performance standards do provide this frame of reference.
Normal distribution of scores described in scale scores, percentiles, stanines, and NCEs.